Skip to content

Benchmark

WildClawBench · Multimodal (26 tasks)

Scores are percentages. Higher is better.

Models and systems measured
38
Scores recorded
38
Sources
1
IQ weight
1.21

Score distribution

41.1%
Mean
40.7%
Median
15.9%
Std dev
23.8%
10th percentile
63%
90th percentile
7.3%
Min
7.3%71%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.21 for chat models, measured over 17 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#Model or systemScore
1Qwen3.8 Omni Flash多模态任务子集 · Claude Code· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.71%
2GPT-5.6 SolOpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs69.3%
3Claude Fable 5OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs68.6%
4Claude Opus 4.8OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs65.3%
5GPT 5.5OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs63%
6Qwen3.8 MaxOpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs59.4%
7Gemini 3.8 Flash多模态任务子集 · Claude Code· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.58.9%
8Claude Opus 4.7OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs58.5%
9Grok 4.5OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs55.3%
10Muse Spark 1.1OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs53.7%
11GLM 5.2OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs50.9%
12Claude Opus 4.6OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs47.7%
13Muse Glimmer 30BOpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs46.6%
14Kimi K3OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs43.8%
15Gemini 3.1 Pro PreviewOpenClaw · multimodal-input split · 26 tasks · low-effort· Reported by WildClawBench / InternLM· via BenchBoard docs43.5%
16Hy3 (Deep Infra)OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs43.1%
17Kimi K2.7 CodeOpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs41.7%
18GLM 5.1OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs40.7%
19GPT 5.4OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs40.2%
20Qwen 3.8 27BOpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs39.1%
21Qwen3.6 27BOpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs38.2%
22MiMo V2.5 ProOpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs36.2%
23Gemma 4 31B ITOpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs34.7%
24Qwen3.5 Omni Plus多模态任务子集 · Claude Code· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.34.5%
25MiMo V2 FlashOpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs34.3%
26DeepSeek V4 Pro 0813OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs33.6%
27GLM 5OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs30.7%
28MiMo V2 ProOpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs29.7%
29DeepSeek V3.2OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs26.1%
30Kimi K2.5OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs24%
31Qwen3.5 397B A17BOpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs23.8%
32MiniMax M2.7OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs19.2%
33MiniMax M2.5OpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs16.3%
34Step 3.5 FlashOpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs12.6%

Not ranked

The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.

Model or systemScore
GLM 5 TurboOpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs24.5%
Grok 4.20 BetaOpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs7.3%
Intern-S2 Preview 397BOpenClaw · multimodal-input split · 26 tasks· Reported by WildClawBench / InternLM· via BenchBoard docs34.2%
Seed-2.0-Lite多模态任务子集 · Claude Code· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.41.9%

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.